Stars
[ACL 2025 Main] Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinions
Code to reproduce the paper "Questioning the Survey Responses of Large Language Models"
open source interpretability platform 🧠
Fully customizable AI chatbot component for your website
The AI framework that adds the engineering to prompt engineering (Python/TS/Ruby/Java/C#/Rust/Go compatible)
💬 A proposal for a web API for prompting browser-provided language models
Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".
A library for mechanistic interpretability of GPT-style language models
Training Sparse Autoencoders on Language Models
Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).
Recipes to train reward model for RLHF.
A recipe for online RLHF and online iterative DPO.
A natural language interface for computers
Modeling, training, eval, and inference code for OLMo
A collection of tools for auditing HAR files and removing personally identifiable data.
An Airtable-like / Excel-like component to create beautiful spreadsheets.